Who is more accurate in the diagnosis of neck of femur fractures, radiologists or orthopaedic trainees?

OBJECTIVE: Our study aims to compare the sensitivity and specificity between orthopaedic trainees and radiologists in diagnosing occult hip fractures from pelvic radiographs.
METHODS: All patients undergoing computed tomography (CT) of the hip for a suspected occult hip fracture at our two centres between November 1, 2010 and October 31, 2011 were retrospectively identified. Two orthopaedic trainees (RJ and ED) independently reviewed the initial radiographs, and if both agreed a fracture was present this was recorded. The reports from the radiologist of the initial radiograph and subsequent CT were recorded, with the CT report used as the standard of reference for comparison of performance.
RESULTS: 239 CTs were performed for suspected neck of femur fractures during the study period, compared to only 20 in 2006-07, reflecting a 1,095 % increase. The interpretation of radiographs by the orthopaedic trainees showed a sensitivity of 55.4 % and specificity of 96.6 %. In comparison, the radiologists had a sensitivity of 60.7 % and specificity of 92.9 %; this difference was shown not to be statistically significant. Radiologists and orthopaedic trainees agreed when interpreting the majority of radiographs with a Kappa co-efficient of 0.77 (a statistical measure of inter-rater agreement). In 20 of the 174 cases (11.5 %) where the radiologist and orthopaedic trainees agreed no fracture was present, the subsequent CT revealed a fracture.
CONCLUSION: Orthopaedic trainees and radiologists have similar accuracy at interpreting pelvic radiographs for suspected hip fractures, so the trainee's opinion can be relied upon. If uncertainty exists, then either the orthopaedic trainee or radiologist should request further imaging urgently as we believe this will expedite diagnosis and treatment of hip fractures, and ensure judicious use of CT.